The EFSA Journal (2005) 236, 1-118

Transcription

1 OPINION OF THE SCIENTIFIC PANEL ON CONTAMINANTS IN THE FOOD CHAIN ON A REQUEST FROM THE EUROPEAN PARLIAMENT RELATED TO THE SAFETY ASSESSMENT OF WILD AND FARMED FISH Question N EFSA-Q Adopted on 22 June 2005 SUMMARY EFSA was requested by the European Parliament to conduct a scientific assessment of the health risks related to human consumption of wild and farmed fish. An EFSA Interpanel working group was set up to conduct this assessment. The opinion focused on the following finfish species as being marketed to a significant amount in the European Union: salmon, herring, anchovies, tuna, mackerel, pilchards, rainbow trout and carp. A special focus was also given to Baltic herring at the request of the European Parliament. Of the selected fish, salmon, rainbow trout and carp are predominantly or exclusively farmed. The other species are predominantly caught from the wild. About two-thirds of fish consumed in the EU is caught from the wild. Species, season, diet, location, lifestage and age have a major impact on both the nutrient and contaminant levels of fish. These levels vary broadly within species and between species in both wild and farmed fish. There is a need for standardisation of sampling procedures before a robust comparison of wild and farmed fish can be made. From the limited data available it seems that if there are any differences between farmed and wild fish, they are small when taking into account the above mentioned factors. However, regional differences exist, e.g. in the Baltic Sea. Contaminants in fish derive predominantly from their diet, and levels of bioaccumulative contaminants are higher in fish that are higher in the food chain. Whilst it is not possible to control the diet of wild fish, the levels of contaminants, and of some nutrients, in farmed fish may be modified by altering their feed. Fish meal and fish oil, are the most important sources of contamination of farmed fish feed with dioxin-like compounds. EU regulations on polychlorinated dibenzo-p-dioxins and furans (PCDD/F) in fish feed were introduced in 2002; the planned inclusion of the dioxin-like polychlorinated biphenyls (DL-PCBs) in the regulations may help to reduce levels of these contaminants in farmed fish. Fatty fish is an important source of long chain n-3 polyunsaturated fatty acids (LC n-3 PUFA). Other substantial natural sources of LC n-3 PUFA are human milk and marine algae. Farmed fish tend to have higher total lipid levels with lower proportions of LC n-3 PUFA than wild fish. Together, these differences mean that the amount of LC n-3 PUFA per portion of fish is similar. Replacement of fish products by vegetable protein and oils in fish feed or decontamination procedures may be a possible means of reducing some contaminant levels. However modification of the fish oil inclusion rate may change the fatty acid composition and in particular reduce the LC n-3 PUFA levels in farmed fish. 1

2 There is evidence that fish consumption, especially of fatty fish (one to two servings a week) benefits the cardiovascular system and is suitable for secondary prevention in manifest coronary heart disease. There may also be benefits in foetal development, but an optimal intake has not been established. Fish can contribute significantly to the dietary exposure to some contaminants, such as methylmercury, persistent organochlorine compounds, brominated flame retardants and organotin compounds. The most important of these are methylmercury and the dioxin-like compounds, for which high level consumers of certain fish may exceed the provisional tolerable weekly intake (PTWI) even without taking into account other sources of dietary exposure. Such exceedance is undesirable and may represent a risk to human health if repeated frequently. However, eating for example meat instead of fish will not necessarily lead to decreased exposure to dioxin-like compounds. Intakes of the other contaminants in fish reviewed in this opinion were not a health concern, because they do not contribute significantly to total dietary exposure and/or it is very unlikely that even high level consumers of fish exceed the health-based guidance values, if available. The greatest susceptibility to the critical contaminants, e.g. methylmercury and the dioxin-like compounds occurs during early development. Exposure during this life stage results from the total amount in the mother s body. For methylmercury it is possible for a woman to decrease the amounts in her body by decreasing intake in the months preceding and during pregnancy, whereas this is not possible for the PCDD/Fs and DL-PCBs because it would take many years to decrease the levels in the body significantly. This evaluation focussed on fish that are widely available in the EU, and likely to be consumed most frequently. Of these, the highest levels of methylmercury are found in tuna, which is mostly caught from the wild. The fish with the highest levels of PCDD/F and DL- PCBs are herring which are caught from the wild and salmon which are mostly farmed. Frequent consumers of Baltic herring and wild Baltic salmon are more likely to exceed the PTWI for PCDD/F and DL-PCBs than other consumers of fatty fish Overall the Panel concluded that with respect to their safety for the consumer there is no difference between wild and farmed fish. Key words: fish contamination, evaluation overview, PCDD/F, PCB, methylmercury, polybrominated flame retardants, quality of feed, wild and farmed fish, Baltic herring, nutritional composition, beneficial effects, fish consumption 2

3 TABLE OF CONTENTS SUMMARY... 1 TABLE OF CONTENTS... 3 ASSESSMENT General introduction Selection of fish species Influence of species, life-stage and season on nutrient and contaminant levels in fish Feeding practices, quality of feed and transfer of contaminants into fish Introduction Transfer of contaminants into fish Dioxins and dioxin-like PCBs (DL-PCBs) Non-dioxin-like polychlorinated biphenyls (NDL-PCB) Polybrominated flame retardants Camphechlor Mercury Cadmium Lead Arsenic Reducing contamination with persistent organic contaminants Conclusions Nutritional Composition of wild and farmed fish General considerations Energy Protein Lipids Vitamins Minerals Nutritional composition of farmed fish Influence of fish farming (aquaculture) Differences in nutritional composition of farmed and wild fish Other differences between wild and farmed fish Conclusion Beneficial effects associated with fish consumption National recommended intakes of fish or LC n-3 PUFA Health benefits from LC n-3 PUFA consumption Risk of cardiovascular disease and stroke Effect of fish or LC n-3 PUFA supplementation in pregnancy on outcome Other beneficial effects associated with fish (fish oil) consumption Adverse effects Summary and conclusions Evaluation overview on fish consumption Toxicity Toxicity of metals Toxicity of selective organochlorine compounds Toxicity of brominated flame retardants Toxicity of organotin compounds Exposure scenarios General considerations Baltic herring

4 Methylmercury Organochlorine compounds Brominated flame retardants Organotin compounds Nutrient intake Risk Characterisation General considerations Methylmercury PCDD/Fs and DL-PCBs Baltic herring and salmon Possible impact of not eating fish CONCLUSIONS AND RECOMMENDATIONS REFERENCES SCIENTIFIC PANEL MEMBERS ACKNOWLEDGEMENT DOCUMENTATION PROVIDED TO EFSA ABBREVIATIONS AND GLOSSARY FOR SOME TERMS APPENDIX ANNEX 1. Tables ANNEX 2. Details of metabolism, function and physiological requirement of LCn-3 PUFA113 4

5 BACKGROUND Consumption of fish is considered to be an important element of a balanced human diet. The aquatic environment, from which it is derived, however, is also the ultimate repository for a considerable range of natural and anthropogenic contaminants. These may be found in the waters themselves, in the sediments, or within the components of the natural web of productivity, and can concentrate in the tissues of fish and shellfish used for human consumption which feed within that ecosystem. Awareness of possible risks associated with such contamination has led to greater controls over release of many contaminants into the ecosystem and levels in the biota are falling for some but not all of these contaminants in many areas. Following a proposal presented by the Commission on 28 August 2001, the Council adopted on 29 November 2001 a Council Regulation amending Commission Regulation (EC) No 466/2001 of 8 March 2001, setting maximum levels for certain contaminants in foodstuffs. This Regulation establishes maximum levels for dioxins and furans in several foodstuffs, including fish and fishery products and products thereof. For a temporary period ending in 2006, the Council granted Finland and Sweden derogation from the application of the maximum levels for wild fish originating from the Baltic region intended for consumption in their territory. The Commission accepted this derogation taking into account that a significant part of the Baltic fatty fish will not comply with the maximum level and would thus be excluded from the Swedish and Finnish diet and such an exclusion of fish from the diet may have negative health impacts. In Sweden and Finland governmental authorities have issued dietary recommendations regarding the consumption of fatty fish from the Baltic and a few big lakes. In Sweden, girls and women of childbearing age are advised not to eat certain fatty fish species from a well defined area more often than once a month. The rest of the population is advised not to eat the same species from the same areas more often than once a week. In Finland children, young people and people at fertile age are advised not to eat large herring or wild salmon from the Baltic more than one or twice a months. Denmark has recently introduced restrictions on the landing of wild salmon over 4.4 kg because of elevated levels of dioxins above EU limits. (Danish national order number 1145 of 25/ ). In the lay press and in a small number of scientific publications, concerns have been expressed that levels of compounds such as dioxins and heavy metals such as mercury represent a health hazard to human consumers even at the current levels found in the tissues of fish. For example, Hites et al. (2004a,b) claimed that consumption of farmed salmon may pose serious risks due to contamination which would detract from any beneficial effects of consuming it. In all of these discussions, the main weight of emphasis has been on the chemical assessment and putative health risk of consumption of wild and farmed fish, but no consideration was given to the nutritional value of fish consumption. TERMS OF REFERENCE The European Parliament requests the EFSA to conduct a scientific assessment of the health risks related to the human consumption of wild and farmed fish (salmon and other carnivorous fish species farmed in substantial quantities) marketed in the European Union. The assessment should focus on the presence and adverse effects in these fish species of persistent organochlorine pollutants (POPs) and other contaminants for which adequate analytical data exists, and on the methodologies for setting safety limits. The European 5

6 Parliament also requests the EFSA that the scientific assessment should cover an overall impact and risk assessment of the consumption of Baltic herring. Interpretation of the terms of reference by the Panels The EFSA notes that methodology for setting limits for contaminants in fish is a risk management function of the European Commission. Therefore the EFSA provides a scientific opinion on the health risks related to the human consumption of wild and farmed fish. The opinion concentrates on some fish species (farmed, wild, marine, freshwater, lean, and oily) marketed to a significant amount in the European Union. The assessment focuses on those chemicals generally considered most relevant in the context of health risks of fish consumption and for which substantial analytical data exist. An EFSA Interpanel working group consisting of members of the Panels on Contaminants in the Food Chain (CONTAM Panel), Dietetic Products, Nutrition and Allergies (NDA Panel), Additives and Products or Substances used in Animal Feed (FEEDAP) and Animal Health and Welfare (AHAW Panel) was set up to prepare this opinion. The scientific opinion addresses: - The influence of season and life history stage of fish on the nutrient levels and the contaminant levels in fish (selecting the right fish comparator) - The quality of feed and feeding practices and its impact on the pattern of contaminants in fish - A comparison of the nutritional composition of wild and farmed fish - The beneficial effects associated with fish consumption - An evaluation of relevant contaminants in fish and comparison with health based guidance values for risk characterisation - An overall impact and risk assessment of the consumption of Baltic herring. Chapter 5 on nutritional composition of wild and farmed fish and chapter 6 on beneficial effects associated with fish consumption were endorsed by the NDA Panel. The opinion was adopted by the CONTAM Panel. ASSESSMENT 1. General introduction Food obtained from the aquatic environment may be of plant (seaweeds) or animal origin. Animals however comprise by far the larger component. They may be representatives of the Invertebrata or the Vertebrata and represent a much greater range of taxonomic diversity than is found in food from terrestrial sources (Bone et al., 1995). They all inhabit a continuous medium and there is some redistribution of energy and anthropogenic contaminants from oceans back to rivers with migratory fish and birds. Generally, however, the flow of energy, and any contaminants, is from land and rivers to the oceans. Although certain forms of shellfish culture and extensive finfish culture have existed for 3,000 years or more, food from the aquatic environment has traditionally been derived from a hunter-gatherer process involving totally wild populations. Recently this process has become increasingly efficient with the advent of high-powered vessels, sophisticated fishing gears and advanced fish finding equipment. As a result, over-fishing has led to a situation where, despite 6

7 increased catching effort, e.g. by deep sea fishing, world total annual catches have been virtually stable or declining since 1986 ranging between 88 and 98 million metric tons (FAO, 2003a). Any consideration of fish consumption in Europe has to take account of the international nature of fish production and trading, since fish is one of the most widely traded commodities. While EU fish catches have been declining, consumption has been increasing by at least 1 % per annum for more than a decade (FAO, 2003a). The difference has been met by increased imports of wild caught fish and, particularly, by the huge increases in both EU derived and imported farmed fish. There is not, however, much direct substitution, as Atlantic salmon, trout and sea bass are the principal products of the expansion of European aquaculture whereas it is the whitefish and in particular the gadoids, that represent the main reduction in production from the capture sector. Production of all species from aquaculture has risen from about 1 million tonnes in 1960 to 46 million tonnes in 2000, the latest date for which statistics are available (FAO, 2003a). At last count, 210 species of finfish, shellfish, molluscs and aquatic plants were cultivated for human consumption (Tacon, 2003). This total included 131 species of finfish, 42 mollusc species, 27 crustacean species, 8 plant species and two amphibian and reptile species. Global finfish production from aquaculture was 23 million tonnes, mollusc production was 11 million tonnes, crustacean production was 1.6 million tonnes and aquatic plant production was 10 million tonnes in 2000 (FAO, 2003a). China is the largest aquaculture producer, at 32 million tonnes, followed by India (12 million tonnes), Japan (1.3 million tonnes), Philippines and Indonesia (1 million tonnes) and Thailand (0.7 million tonnes) (FAO, 2003a). Norway is the largest salmon producer (488,000 tonnes), with the US and Chile not far behind (428,000 and 425,000 tonnes). US production of Pacific salmon, included in this figure, however, is a mix of wild, farmed and ranched production. Traditionally in Western Europe, marine species such as cod, haddock, mackerel and herring, and canned Pacific salmon, have been consumed in significant quantities, with fish such as the Atlantic salmon and rainbow trout being much less available. In Mediterranean countries, a wider range of species including sea bass, bream and tunas, have been consumed, while in Germany and eastern countries of Europe, freshwater carp from pond culture has traditionally been of significance. Imports of frozen or canned fish have also been considerable, particularly tunas, white fish such as hake and cod and Pacific salmon. The importance of fish and in particular the fatty fishes such as herring, mackerel, tuna and salmon, as a dietary source of long chain n-3 polyunsaturated fatty acids (LC n-3 PUFA), is well recognised. Unfortunately it is also within the lipid component of the fish that lipidsoluble contaminants such as dioxins and polychlorinated biphenyls (PCBs) are stored. Fish is a generic term and there are wide differences in both nutritional value and in potential contaminant levels, depending not only on the origin and the fish species, but also on the tissue sampled, the season of harvest and, for farmed fish, the content of the diet. The diet of wild fish is totally beyond human control (apart from global measures to decrease the release into the environment) and it is only with the development of formulated diets, for farmed species such as salmon and sea bass that the opportunity to directly control tissue contaminant levels has become available. Such considerations are essential in relation to any assessment of comparative contaminant levels and their significance. Comparisons between farmed and wild fish are particularly difficult in this context, and it is essential to compare like with like. Hites et al. (2004a,b) for example compared farmed Atlantic salmon (Salmo salar) harvested in mid reproductive cycle from Atlantic Oceanic waters and wild Pacific salmon (Oncorhynchus spp), of different species, captured in pre-spawning condition, from Pacific coastal waters. 7

8 The inter-relationship between lipid levels and any possibly harmful contaminant levels in fish is one which has to be addressed in any dietary recommendations. Such recommendations require understanding of the factors involved and the comparative risks of recommending an increase or reduction in fish consumption. This assessment focuses on selected wild and farmed fish species consumed in Europe (chapter 2). Chapter 3 describes the factors that may influence the levels of contaminants in fish, which would need to be standardised to allow a robust comparison of nutrients and contaminants in wild and farmed fish. Chapter 4 addresses the contribution of diet to levels of contaminants in and the opportunities for reducing contaminant levels in farmed fish. The assessment also reviews the nutrients in fish (chapter 5) and the specific beneficial effects of eating fish (chapter 6). A wide variety of contaminants may be present in fish. In particular, there are numerous lipophilic organochlorine contaminants in the environment including polychlorinated dibenzo-p-dioxins and furans (PCDD/F), polychlorinated biphenyls (PCBs), camphechlor, hexachlorocyclohexane, dichlorodiphenyltrichloroethane (DDT) and its metabolites (DDD, DDE), chlordane, dieldrin, aldrin, endrin, heptachlor and hexachlorbenzene. Most of these compounds are no longer produced, levels in the environment are generally decreasing and they tend to occur in parallel. It is beyond the scope of this opinion to review all of them in detail, or to evaluate the toxicological data. The assessment therefore focuses on some examples of heavy metals and lipophilic contaminants for which adequate data are available and for which previously established health-based guidance values may be exceeded by some consumers (chapter 7). 2. Selection of fish species The amount of fish on the EU market was used as the criterion for the selection of relevant fish species to be considered in this opinion. This was based on the statistics from FAO (FAO, 2003a), see Table 1. It should be noted that calculations give an overall picture of the landings but cannot take into account local situations in which a group of consumers would eat a single species not considered as significant at the EU level or which would be imported from third countries. 8

9 Table 1. Landings of Fish Species in EU (FAO, 2003a) ( Fish species Catches/produced Tonnes/year Production/inhabitant** kg/year Salmon 674,000* 1.50 Herring 651, Anchovy 387, Tuna 353, Mackerel 357, Pilchard 336, Rainbow trout 227, Carp 131, Cod 76, Sea bream 71, Haddock 53, Sea bass 48, Perch 13, Pike 11, Eel 6, * For salmon the figure includes Norway ** Assuming 450 million inhabitants According to the terms of references, species produced in significant amounts in the European Union should be considered in this opinion. For practical reasons this was defined as species produced in amounts higher than 100,000 tonnes/year. In consequence the selected fish species for consideration in this opinion are rainbow trout, salmon, tuna, herring, mackerel, pilchard anchovies and carp. However the Panel noted that on a national level some other species such as cod, seabream, haddock and seabass may also be consumed in significant amounts. Nevertheless, the list of those species considered significant for this opinion, includes three which are almost entirely produced by culture (Atlantic salmon, rainbow trout and common carp). The remainder are derived from commercial fisheries. Fish species are often categorized into carnivore, omnivore, detritivore and herbivore (De Silva and Anderson, 1995). This has an influence on the concentration of contaminants which derive from the food chain. Of the eight species compared in this study, four (salmon, trout, tuna and mackerel) are carnivores, at the top of the food chain; the others would generally be classified as omnivores. Table 1 of Annex 1 lists common and Latin names of some fish species. The panel noted that from the available statistics it would appear that about two thirds of the fish consumed in Europe is from wild sources and one third is farmed. 3. Influence of species, life-stage and season on nutrient and contaminant levels in fish Comparing chemical contaminant levels in fish is complicated by the variation that is induced by the age of the fish, the geographic origin and the season the fish is harvested or caught. In the wild older fish are generally larger, and will eat larger prey species. Thus they have the opportunity to accumulate higher levels of contaminants over a longer period than their 9

10 younger, smaller peers in the same population. Influences of the fish species, environment and where relevant, the farming system, on possible contaminant levels are many and varied. It is generally accepted that most contaminants are derived from the aquatic food chain rather than directly from the water. Because storage lipids, the main repository tissue for lipophilic contaminants in fish, vary widely with season and life stage, this leads to significant inconsistencies in recorded level of contaminants depending on species, age and tissue sampled. For example, all species store lipid prior to maturation and subsequently transfer maternal nutrients to developing ovaries. Consequently, fish captured or harvested in the early stages of maturation will have high lipid contents in tissues and organs, whereas those captured or harvested after spawning will have low lipid contents in tissues and organs. Cod are very fatty in the spring and summer months when plankton and/or prey levels are high, and most of this fat is stored in the liver. In winter they have lower levels of liver lipid. Salmon on the other hand store most fat in the abdominal peritoneal lipoid tissue, the intermyotomal fascia and particularly in the dermis of the skin, not in the liver. Mackerel and tuna, other oily fish, are particularly likely to store fat in skeletal muscle fascia. Thus the lipid and contaminant levels to be recorded from fish are critically dependent on which tissue is sampled and generally at what time of year. Differences in life stage also complicate comparisons between farmed and wild salmon contaminant levels for the same species. Farmed salmon are generally harvested well before reaching sexual maturity, whereas wild salmon are generally captured just before ascending rivers to spawn. At this stage wild fish have ceased to eat and much of their lipid will already have been transferred from muscle fascia to gonad. This will influence the comparison between levels of contaminants in edible tissues of farmed and wild fish. Among salmonids the situation is also dependent on the species of salmon. For example, there are five species of Pacific salmon that are captured or farmed in significant numbers. Chinook and coho salmon are farmed as well as being captured as wild fish, whereas sockeye, pink and chum salmon are only captured (although a proportion of these are sea-ranched in Alaska). The five species of Pacific salmon have different life histories and also exploit different ecological niches and target different trophic levels and species as their principal prey. Sockeye salmon are planktivores, whereas chinook and coho salmon are carnivorous, consuming mainly small forage fish along with squid or shrimp (krill) (Groot et al., 1995). Pink and chum salmon are somewhat between sockeye and the top predator salmon species in terms of target prey. Interpretation of comparisons of contaminant levels found in farmed and wild salmon should take this into account. Some of the Pacific species which have been claimed to have lower contaminant levels than farmed Atlantic salmon (Hites et al., 2004a,b) are omnivores, and should not be compared with carnivores at the top of the food chain. Thus there is clearly a need for standardization of sampling procedures as well as analytical protocols if any valid comparisons of contaminant levels are to be made between different fish species and between fish derived from farmed and wild sources. Aquaculture Finfish aquaculture may be carried out in cages in fresh water lakes or in ponds, tanks or raceway systems supplied by ground water (springs or wells) or surface water (reservoirs or rivers). Some freshwater facilities re-use water (recirculation systems) and many have sophisticated water filters, heating systems and waste water effluent controls. Such systems are only suitable for producing high-value products, such as salmon smolts or exotic species for specialized live markets. Aquaculture is also carried out in seawater, often called 10

11 mariculture, where fish are often hatched in sophisticated pumped seawater hatcheries, on land, but grown on in cages in the sea. Because of the limits of freshwater availability it is likely that any major future expansion of aquaculture will be in the sea. Aquaculture, irrespective of medium, is of three basic types: extensive, semi-intensive and intensive. Extensive aquaculture, where fish are held in low density earthen ponds and expected to forage for themselves, with possibly some fertilization of the pond to enhance eutrophication, was a long-standing method of production of considerable tonnages of carp in China and also Eastern Europe. Production from extensive systems typically ranges from kg/hectare/crop. Cost of land and limited sources of water have, however, driven production towards semi-intensive culture, where feed is given to supplement the natural feeding and, increasingly, towards intensive production, where all of the nutrients required for growth are provided by means of an extraneous formulated diet. Production from semiintensive systems ranges from 1,000-3,000 kg/hectare/crop, whereas in intensive production, yields of 5,000-10,000 kg/hectare/crop can be achieved. Higher inputs increase the cost of production, but proportionally higher yields make such systems economical to operate. Another form of aquaculture practised in some regions is sea ranching. This makes use of the homing instinct of fish such as the salmonids. Very large numbers of fry are reared intensively in hatcheries and released into the sea, where they feed on their natural prey species. They may then be captured at sea, as happens in the Baltic, where almost all fish are originally derived from hatcheries, or when they return to the estuary of their home river, as in the case of Alaskan salmon, some 30 % of which are hatchery derived. Sea ranching is a relatively low-input form of aquaculture, but only suitable for homing species or those that remain close to the vicinity of their release point. In mariculture, the largest tonnages of finfish in production terms are Atlantic salmon (0.9 million tonnes) produced principally in Norway, Chile, Scotland, Ireland and Canada. Japanese amberjack is the next largest marine production species at 137,000 tonnes. Sea bass/sea bream produced in Spain, France, Italy, Greece and tuna produced in Australia, Croatia and Malta are growing but still small mariculture crops compared to production of established species. Newer species now in culture include cod, halibut, turbot and barramundi. All mariculture species are produced in floating cages at sea, although the Atlantic salmon has to have a fresh water stage of 6-12 months in fresh water before going to sea. 4. Feeding practices, quality of feed and transfer of contaminants into fish 4.1. Introduction In 2000, an estimated 15 million tonnes of feedstuffs were produced for aquaculture, using approximately 2.4 million tonnes of fish meal and 550,000 tonnes of fish oil produced from marine forage species not used to supply seafood for human consumption. Approximately one-third of all seafood landings are used to make fish meal and fish oil, with the remaining two-thirds consumed directly. All of the key economic species farmed for the Western market are top trophic-level predators which in their major growth phases, hunt and consume prey species, principally finfish and crustaceans. They derive metabolic energy from protein metabolism and consequently carbohydrate plays little part in their energy supply (Halver and Hardy, 2002). They also require significant levels of certain PUFAs. In the wild, they acquire these from the prey species that concentrate them from marine plankton but they must be supplied in farmed fish 11

12 diets, generally from fish oils. Salmonids, e.g., Atlantic salmon, Pacific salmon, trout and char, are an exception to this requirement as they can to some degree synthesise their own longer chain PUFAs from 18-carbon, n-3 precursors found in plant oils. This ability to synthesise LC n-3 PUFA is severely reduced in wild fish following migration to seawater. Tuna farming at present depends on capturing young tuna at sea and holding them in very large cages off shore for a year or more. They are fed by-catch species and do not currently receive formulated diets. The contribution of different feedstuffs to compound feed varies broadly with the category of fish because of the species dependent adaptation of digestive systems to dietary substrates like starch, crude fibre and fat. In general, the majority of relevant fish species needs mixed diets, but varying ingredients in the compound feed. Table 2 lists typical diets for omnivorous (emphasizing plant feedstuffs) and carnivorous fish species. Table 2: Typical composition of omnivorous and carnivorous fish diets (EC SCAN, 2000) Feed materials Omnivorous fish Carnivorous fish (fish diet expressed in % ) Cereals (wheat, corn) Oilseed meals (soybean meal) 56 7 Corn gluten meal - 5 Fish meal Fish oil 2 25 Premix* 2 2 * Includes minerals, trace elements, vitamins, single cell proteins and other feed additives 4.2 Transfer of contaminants into fish Diet is a main source of exposure to a wide range of contaminants in fish, although uptake also occurs via the gills. In farmed fish the level of contaminants in feed materials can be monitored and controlled, whereas in wild fish exposure remains unknown and will vary considerably in different geographical regions Dioxins and dioxin-like PCBs (DL-PCBs) The concentrations of persistent organic contaminants are highly variable as demonstrated for PCDD/Fs and DL-PCBs in Table 3 and depend primarily on the inclusion level and type of lipid in the diet. 12

13 Table 3. Variation of dioxin (PCDD/F), dioxin-like PCB (DL-PCB) and total (PCDD/F + DL-PCB) contamination of selected feedstuffs in fish nutrition (ng WHO-TEQ/kg 12% moisture content) for different feedingstuffs sampled between 1997 and 2004 (upperbound concentrations) (adapted from Gallani et al., 2004). FEEDINGSTUFFS N DIOXINS + FURANS DIOXIN-LIKE PCBs TOTAL TEQ Feed materials of plant origin Low 5 th %ile Median High 97.5 th %ile Low 5 th %ile Median High 97.5 th %ile Low 5 th %ile Median High 97.5 th %ile Vegetable oils Premixes* Animal fat, incl. milk and egg fat Fish oil Fish and other aquatic animals and their products (fish meal) Feedingstuffs for fish * Two data have been removed from the database as they are clearly to be considered as outliers. 13

14 Data (EC SCAN, 2000) indicate that fish oil and fish meal from European production contain higher levels of PCDD/F and DL-PCBs than fish oil and fish meal of the South Pacific origin. In omnivorous diets the contribution of fish products to the total contamination with PCDD/F and DL-PCBs was above 55 % and may be up to 98 % in carnivorous diets (EC SCAN, 2000) The final feed will comply with the actual regulatory limit (2.25 ng/kg) if the individual components also comply with their respective limits (fish meal, 1.25 ng/kg; fish oil, 6 ng/kg). Karl et al. (2003) reported data on the transfer of PCDD/F from commercial fish feed produced in Norway into the edible part of rainbow trout (Oncorhynchus mykiss). In muscle tissue PCDD/F increased continuously up to the end of the experimental period (0.914 ng PCDD/F WHO-TEQ/kg meat). The mean transfer rate ranged from 11.1 % at 6 months to 30.7 % at 19 months. The transfer rate in females appeared lower possibly as a result of distribution into the eggs. A direct correlation (r 2 = 0.98) between concentration in the lipid fraction of feed and in fish was observed. The transfer rate for DL-PCBs was shown to be higher than that of PCDD/F in Atlantic salmon (Isosaari et al., 2004; Lundebye et al., 2004) and rainbow trout (Isosaari et al., 2002). Table 4 shows dioxin and DL-PCBs contamination matter of Atlantic salmon due to feeding differently contaminated feed. Considerable variation was observed in transfer rates of dioxins (40 65 %) and DL-PCBs (78 93 %) because of different transfer of the congeners from feed to fillet in Atlantic salmon over an entire production cycle (Berntssen et al., 2005). Among PCDD/Fs, tetra- and pentachlorinated congeners were found to be preferentially accumulated in salmon, while hepta- and octachlorinated dibenzo-p-dioxins were excreted into the feces. Congener patterns that were associated with a preferential accumulation of PCBs in salmon included non-ortho and tetrachloro congeners. Non-ortho tetrachloro congeners were preferentially accumulated compared with other non-ortho and mono-ortho PCBs (Isosaari et al., 2004). Table 4. Dioxin and DL-PCBs in Atlantic salmon following feeding of differently contaminated mixed feed (A-D) over 30 weeks (Lundebye et al., 2004) Diet A B C D Dietary contaminants from different fish oil sources (pg WHO-TEQ/g dry matter) PCDD/F DL-PCBs Total Contaminants in whole fish* PCDD/F (pg WHO-TEQ/g wet weight) DL-PCBs Total * At the end of the experimental period (30 weeks) A model calculation based on different assumed transfer rates, indicates that even with a very high transfer rate (80 %), the total PCDD/F can be expected to be about 50 % of the current maximum permitted EU level for PCDD/F in fish, i.e. 4 pg WHO/TEQ/g wet weight (Commission Regulation (EC) No 466/2001). There is only a limited database for a reliable comparison of wild and farmed fish and data on the related dietary contamination are mostly not available. A recent overview on dioxins and DL-PCBs in fish in the EU is provided in Table 5, but this does not allow a comparison of wild and farmed fish as it includes only very partially the same species. The farmed fish is mainly dominated by salmon and trout, whilst the wild fish 14

15 covers a wider range of fish species, and wild salmon and trout are very minor contributors in the data base of wild fish. The Swedish Food Administration is monitoring basic data regarding the concentration of non-readily biodegradable environmental organic contaminants in fatty fish from Sweden (Baltic Sea, lakes Vänern and Vättern, waters along Sweden west coast). PCDD/F levels in fatty fish from Sweden including wild and farmed fish species are provided in Table 2 of Annex 1. However conclusions based on comparison of these results should be made with care because of differences in size of fish and the season and location that they were caught. 15

16 Table 5. Occurrence data on dioxins, furans and dioxin-like PCBs in food based on Gallani et al Data are expressed as ng WHO-TEQ/kg fresh weight FOOD N DIOXINS + FURANS DIOXIN-LIKE PCBs TOTAL TEQ SAMPLES Average Median 90 th %ile 95 th %ile Average Median 90 th %ile 95 th %ile Average Median 90 th %ile 95 th %ile 99 th %ile FISH Total wild + (426) farmed except fish from the Baltic region - wild (215) farmed (211) Herring (53) Fish other than herring (373) BALTIC FISH (total) (340) Baltic herring (173) Baltic processed (14) herring - Baltic pike-perch (39) Baltic salmon (22) Baltic other 2 (89) Herring from other regions compared to herring from Baltic region - Herring other regions (53) Baltic herring (173) Based on data contained in the report Dioxins and PCBs in Food and Feed: Data available to DG SANCO Joint Report DG SANCO / DG-JRC-IRMM and data submitted after 31 January 2004 by the EU-Member States (only data from 1998/1999 onwards have been taken into account. NB: farmed fish is mainly dominated by salmon and trout; wild fish covers a wider range of fish species with few wild salmon and trout. 2 Burbot, sprat, bream, white fish, roach, vendace, flounder, smelt, eel, brown trout, cod. The three samples of River lamprey are not included (high levels). 16 of 118

17 Non-dioxin-like polychlorinated biphenyls (NDL-PCB) Data on the level of contamination of fish feed with NDL-PCB (expressed as the sum of the 6 indicator PCB) presented in table 6 show that mean concentrations vary from 10.7 to 54.7 ng/g depending on the type of feed product. Table 6. NDL-PCB (Σ 6) contamination of selected feeding stuffs in fish nutrition (ng/g product) Feedingstuff N Mean Median Min Max Feed material of plant origin Fish oil Fish and fishery products Feedingstuffs for fish Based on the information presented in the opinion of the Scientific Panel on Contaminants in the Food Chain on the Presence of NDL-PCB in Food and Feed (EFSA 2005, in preparation) it can be concluded that transfer of NDL-PCB from fish feed into fish is comparable to that in other animal species. Higher chlorinated indicator PCB (PCB 138, 153, 150) shows a greater transfer than the lower chlorinated congeners. Just as for dioxins and DL-PCB, also for NDL-PCB there is only limited information allowing a comparison between levels in farmed and wild fish. An overview on levels of NDL-PCB in fish from EU Member States is given in table 7, showing particularly high levels of NDL-PCB in Baltic fish. Table 7. Occurrence of NDL-PCB (Σ 6) in fish and fish products in ng/g fish. Fish N Mean Median Min Max Fish and fishery products (including Baltic fish) Baltic fish all fish - herring salmon Polybrominated flame retardants Recent market basket studies have detected polybrominated diphenyl ethers (PBDEs) in a wide range of food, including fish and other seafood species, and national food surveys have identified the diet as one of the main sources of human exposure to these brominated flame retardants (BFRs) (Bocio et al., 2003; Ohta et al., 2002; Schecter et al., 2004). 17 of 118

18 Information related to fish feed contamination with PBDEs is limited (Bethune et al., 2005, Table 3 of Annex 1; see also Dietary accumulation of PBDEs has been investigated in feeding trials with Atlantic salmon (Isosaari et al., 2005), zebra fish (Andersson et al., 1999), juvenile common carp (Stapleton et al., 2002, 2004a,b,c), pike (Burreau et al., 1997, 2000), juvenile rainbow trout (Kierkegaard et al ) and juvenile lake trout (Tomy et al., 2004). A wide range of congener-dependent accumulation was reported, ranging from less than 0.02 to 5.2 % for BDE 209 (Kierkegaard et al., 1999; Stapleton et al., 2002) to more than 90 % for BDE 47 (Burreau et al., 1997, 2000; Stapleton et al., 2004b). Isosaari (2005) found that Atlantic salmon accumulated PBDEs from feed as efficiently as non-ortho PCBs and more efficiently than other PCBs and PCDD/Fs (Isosaari et al., 2004). Biotransformation and/or preferential accumulation of certain PBDE congeners led to differences in the congener patterns between feeds and fish (Isosaari et al., 2005). Summarized data from different studies on the occurrence of PBDEs in fish samples (Table 4 of Annex 1) show considerable variation in concentrations. PBDE concentrations up to 3 4 ng/g fresh weight have been reported in salmon fillets (with skin), and concentrations ranging from 0.49 to 10.9 ng/g wet weight have been found in fish feed (Hites et al., 2004b). Farmed European Atlantic salmon were found to have the highest mean PBDE concentration and wild Pacific salmon had the lowest, (Hites et al., 2004b). Similarly a study of 16 PBDE congeners in several fish species conducted by the FSAI (documentation provided to EFSA) found that farmed Atlantic salmon had the highest PBDE concentration of the fish species examined (mean 3.71 ng/g fish; range ng/g fish), considerably higher than found in wild salmon (mean 0.86 ng/g fish; range ng/g fish). Data from Sweden on wild Baltic salmon (ten pooled samples containing five fishes each) reveal levels similar to those found in farmed Irish Atlantic salmon (mean 2.95 ng/g fish; range ng/g fish). However, these and other available data on PBDE concentrations in farmed and wild salmon (Haglund et al., 1997; Manchester-Neesvig et al., 2001; Easton et al., 2002; Jacobs et al., 2002; Bethune et al., 2005) show that there is a large variation among salmon species, sampling locations and salmon individuals, with no consistent differences between wild and farmed salmon Camphechlor Limited data on transfer rates from one study of three camphechlor congeners (#26, 50 and 62, Parlar nomenclature) from feed to rainbow trout are summarized in Table 5 of Annex 1. Average transfer rates seemed to be about 25 % (Karl et al., 2002) Mercury The retention of dietary mercury by fish is dependent on dietary concentration and exposure duration. For example, Lock (1975) found that the retention of mercury from feed by rainbow trout decreased from 71 % to 38 % following exposure to 3.4 mg Hg/kg feed and 21.6 mg Hg/kg feed for 1 week respectively. Furthermore, accumulation decreased from 71 % to 31 % following 12 weeks exposure of Atlantic cod (Gadus morhua) to 3.4 mg Hg/kg. Mercury from feed appears to have a specific affinity for muscle tissue in fish. (Julshamn et al., 1982). The maximum permitted level of mercury in fishery products in the EU is 0.5 mg/kg wet weight for most species, and 1 mg/kg for a limited list of fish species including tuna (EU, 2001). For fish feed the maximum permitted level for mercury is 0.1 mg/kg (European Community, 2002). Maage et al. (2005) reported mean mercury contents for 5 sampling time points between between mg/kg wet weight in farmed Atlantic salmon (Salmo salar) (range , n = 225). Average concentrations and ranges for methylmercury in 4 different tuna species are given in table of 118

19 Table 8. MeHg values (in µg/g fish) in different tuna species (French Authorities, DGAL, DGCCRF, IFREMER, FIAC) Tuna Species Skipjack Albacore Yellowfin Bluefin Number of samples Average Min ND 0.26 Max ND: Not Detected Cadmium Cadmium accumulates primarily in the viscera (intestine, liver and kidney) of fish (Kraal et al., 1995; Berntssen et al., 2001). Transfer of dietary Cd into fish muscle is low (2 6 %) (Cincier et al., 1998). High dietary exposure of salmon to Cd concentrations up to 250 mg/kg for four months caused significant accumulation in several tissues including muscle (up to approximately 0.25 mg/kg), whereas there was no significant accumulation at dietary concentrations up 5 mg/kg (Berntssen et al., 2001). The EU maximum level for cadmium in fish feed is 0.5 mg/kg (European Community, 2002). The cadmium content reported for commercial salmon fillets (n = 225) is less than mg/kg (Maage et al., 2005), whereas the maximum permitted level of cadmium in fish fillets is 0.05 mg/kg for most species, and 0.1 mg/kg for certain species of fish (EC No 466/2001) Lead Although lead exists in many different forms in marine and fresh waters, most of the lead found in fish is inorganic in nature, and is bound to proteins. Bioaccumulation of lead in marine animals is low compared to mercury (Dietz et al., 1996). Internal organs, and especially skin and bone appear to be the main sites of lead accumulation in fish, whereas lead does not appear to accumulate in fish muscle tissue (Somero et al., 1977). The lead content reported for commercial salmon fillets is less than 0.01 mg/kg, (Maage et al., 2005, n = 225), whereas the maximum permitted level of lead in fish fillets is 0.2 mg/kg for most fish species, and 0.4 mg/kg for selected species (EC No 466/2001). The EU maximum level for lead in fish feed is 5 mg/kg (European Community, 2002) Arsenic Marine organisms accumulate arsenic, predominantly as non-toxic arsenobetaine and arsenocholine. Products like fishmeal and fish oil have been identified as major sources of feed contamination with arsenic, and it is likely that the measured arsenic represents predominantly these organic compounds. The EU maximum level for arsenic in fish feed is 6 mg/kg and for inorganic arsenic is 2 mg/kg. (European Community, 2002) Reducing levels of persistent organic contaminants Different strategies in fish feeding are under investigation to reduce fish meal and fish oil in farmed fish diets, mainly due to limited availability of fish oils. The possibilities and limitations of such a substitution strategy are mainly related to different fish species and depend on physiological needs. 19 of 118

20 Between 25 and 90 % of the fish meal in fish feed can be replaced by plant proteins (Lim, 2004) depending on the target fish species, but plant feedstuffs have lower nutrient concentration, lower digestibility and occurrence of certain natural constituents, such as inhibitors of proteolytic enzymes, isoflavones or glucosinolates (Mambrini et al., 1999; Burel et al., 2000). Furthermore, a high concentration of plant phytates, primarily reducing mineral bioavailability in fish diets, is an important limiting factor for the maximum inclusion of plant feedstuffs (Liebert and Portz, 2005). Substitution of fish oil by vegetable oil in fish feed significantly reduced PCDD/F and DL- PCBs concentration in farmed salmon, Salmo salar (Bell et al., 2004; Berntssen et al., 2005). Complete replacement of fish oil by a mixture of vegetable oils (55 % rapeseed, 30 % palm, 15 % linseed) did not lead to any adverse effects on growth, total muscle lipid content or organoleptic properties in rainbow trout (Corraze et al., 2004). Dioxin levels were twenty times lower in salmon fillet following fish oil substitution by vegetable oil (Berntssen et al., 2005). However, these studies indicate significant changes in the n3/n6 fatty acid ratio in fish products due to this substitution, which may have implications for the nutritional benefits of fish consumption. Genetic selection procedures are currently under way to increase the capacity in farmed fish to synthesize longer chain PUFAs from C 18, n-3-precursors found in plant oils. Another possibility for reducing contaminant levels in feed is the selective use of marine feed ingredients with relatively low natural levels of persistent organic contaminants as previously mentioned. Using certain marine fish oils such as oil from Pacific Ocean species in fish feed can lead to reduced levels of dioxins and to a lesser degree dioxin-like PCBs in farmed Atlantic salmon (Isosaari et al., 2004; Lundebye et al., 2004). A final option for reducing contaminant levels in feed is the technical removal of contaminants from fish meal and fish oil, e.g. by activated carbon treatment, combined with stripping at low pressure and low temperature (De Kock et al., 2004) or short path distillation (Breivik and Thorstad, 2004). However, practicality of different procedures needs to be demonstrated as well as the efficiency for the removal of different contaminants/congeners and maintenance of the nutritional quality of the oil Conclusions The level of contaminants in fish is related to the concentration in the diet, and the duration of exposure. Therefore the level of contaminants in carnivorous species irrespective of whether wild or farmed, may be higher compared to omnivorous species. In farmed fish, marine feed ingredients, primarily fish oil and to a lesser extent fish meal, are the main sources of organic contaminants. Maximum permitted levels are set for a range of contaminants in animal feed in the EU (Directive 2002/32/EC). Research and monitoring programmes indicate that these levels ensure that the concentrations of undesirable substances in fish are below maximum levels permitted for these contaminants in food, where such regulations exist. Risk assessments of undesirable substances in animal feed have recently been conducted by EFSA (EFSA, 2004c,d; EFSA, 2005a,b) and are still on-going. The replacement of fish products by plant feedstuffs in fish diets or decontamination procedures may be a possible means of reducing some contaminant levels. Lowering of fish oil inclusion rates of farmed salmon and trout diets could significantly reduce flesh lipid content and hence contaminant levels but this would also change the fatty acid composition of the fish. 20 of 118